it’s raining men! hallelujah? - hec lausanne! 1! it’s raining men! hallelujah? pauline grosjean...
TRANSCRIPT
1
It’s Raining Men! Hallelujah?
Pauline Grosjean and Rose Khattar*
30 November 2015
Abstract
We document the implications of missing women in the short and long run. We exploit a
natural historical experiment, which sent large numbers of male convicts and far fewer
female convicts to Australia in the 18th and 19th century. In more male-biased areas,
women historically married more and were less likely to work. Today, in areas that were
more male-biased historically, people have more conservative attitudes towards women
working, women are less likely to have high-ranking occupations, and women earn a
lower wage income. We document the role of vertical cultural transmission and of
marriage homogamy in sustaining cultural persistence.
Keywords: Culture, gender roles, sex ratio, natural experiment, Australia
JEL codes: I31, N37, J16, Z13
* University of New South Wales. Corresponding author: Pauline Grosjean, School of Economics, UNSW, Sydney NSW 2052, Australia. [email protected]. We are grateful to Renee Adams, Chris Bidner, Alison Booth, Monique Borgerhoff-Mulder, Sam Bowles, Rob Boyd, Rob Brooks, Raquel Fernández, Gigi Foster, Gabriele Gratton, Ralph de Haas, Kim Hill, Hongyi Li, Leslie Martin, Terra McKinnish, Suresh Naidu, Allison Raith, Marc Sangnier, Paul Seabright, Jesse Shapiro, Peter Siminski, and audiences at the NBER Summer Institute Political Economy 2014, the Santa Fe Institute, Australian National University, the Institute for Advanced Studies in Toulouse, Hong Kong UST, Monash University, QUT, Sciences-Po, University of Sydney, UNSW and Wollongong University for comments and suggestions. This research has benefited from the financial support of UNSW Business School. This paper uses unit record data from the Household, Income and Labor Dynamics in Australia (HILDA) Survey. The HILDA Project was initiated and is funded by the Australian Government Department of Families, Housing, Community Services and Indigenous Affairs (FaHCSIA) and is managed by the Melbourne Institute of Applied Economic and Social Research (Melbourne Institute). The findings and views reported in this paper, however, are those of the authors and should not be attributed to either FaHCSIA or the Melbourne Institute. All errors and omissions are the authors’.
2
1. Introduction
Despite some improvements in the last century, gender disparities are persistent both in
the workplace, where women work less, earn less, and do not reach the top positions; and
at home, where women still do more than their fair share of housework (Bertrand et al.
2013). Gender-based discrimination manifests most severely in regions where highly
skewed, male-biased sex ratios prevail: there is now an estimated 80 million women
“missing” in China and India alone, due to sex-selective abortion and differential gender
mortality (Hesketh and Xing 2006). An important question is how male-biased sex ratios
further affect female outcomes. Answering this question is very difficult because of
endogeneity issues, since a surplus of men is itself the product of lower opportunities for
women (Qian 2008, Carranza 2014) and of cultural preferences for sons (Almond et al.
2013). The ideal natural experiment would consist of dropping a larger number of men
than women on an isolated island, with these men and women being of a similar cultural
background and operating in the same institutional environment, and then observe female
outcomes from that point on.
We exploit such an experiment. Heavily male-biased sex ratios resulted from the British
policy of sending convicts to Australia in the late 18th and 19th centuries. Men far
outnumbered women among convicts, by a ratio of 6 to 1 (Oxley 1996). The vast
majority of the white Australian population initially consisted of convicts.1 Even among
free migrants, men vastly outnumbered women well into the 20th century, as mostly men
were seeking out Australia’s economic opportunities in mining and pastoralism. As can
be seen in Figure 1, a male-biased sex ratio endured in Australia for more than a century.
We rely on spatial and time variation in the historical sex ratio and study the short- and
long-term effects of male-biased sex ratios on female outcomes at home and in the
workplace. We find that historically, gender imbalance was associated with women
marrying more, participating less in the labor force, and being less likely to work in high-
ranking occupations. We then study the long-term implications by matching, for the first
time, 91 historical counties from the Australian Colonial Censuses to postal areas in the
1933 Census, the 2011 Census, and in a nationally representative household survey 1 There is still uncertainty and controversy about the number of Aborigines at the start of European settlement but in any case, they constituted a very different economy.
3
collected between 2001 and 2011. In areas that were more male biased historically,
people today have more conservative attitudes towards women working, women are less
likely to have high-ranking occupations, they work less, and as a result, they receive
lower wage income. However, we find no evidence that the overall effect on women’s
welfare is negative, as measured by self-reported marital and overall life satisfaction. The
“glass ceiling” effect was already present, and larger in magnitude in 1933, before the
onset of large migration flows to Australia after the Second World War.
In terms of magnitude, the legacy of historical sex ratios on attitudes towards gender
roles is comparable to the effect of urbanization. Moreover, a one standard deviation
increase in the historical sex ratio is associated with a decrease in the share of women
employed as professionals by 0.13 standard deviations. Historical circumstances explain
more than 3% of the variation in the share of women employed as professionals that is
left unexplained by traditional factors, even when accounting for the share of men
employed in similar professions.
A concern for identification is that gender imbalance across localities was determined by
characteristics that also influence the outcomes of interest. Our historical results are
robust in a panel of historical counties, controlling for time and county fixed effects.
These remove the influence of any time invariant county characteristics that could be
associated both with local gender imbalance and with female work outcomes. Results
pertaining to the legacy of the sex ratio on present day outcomes are only available in a
cross section, so we cannot pursue a similar strategy. Instead, we start by carefully
analyzing the characteristics that determined the allocation of men and women across
counties and control for such factors in the regression analysis. Historically, economic
opportunities consisted chiefly of agriculture, pastoralism and mining. To account for
these, we flexibly control for geographic characteristics by including latitude and
longitude in all specifications and control precisely for the presence of minerals and for
terrain characteristics, as well as for the initial conditions in the county in terms of
economic specialization. We also account for a wide range of present day controls. State
fixed effects are used throughout in order to remove any unobserved heterogeneity
related to differences in the legal environment or in the treatment of convicts.
4
Although we are able to account for the influence of a large number of geographic and
historical characteristics, it remains possible that local gender imbalance in the past was
influenced by unobservable characteristics that still underlie female opportunities as well
as attitudes in the present. Endogeneity could arise, for example, from the systematic
selection of women with stronger preferences for leisure or of men with a taste for gender
discrimination to high historical sex ratio areas. To deal with this, we employ an
instrumental variable strategy based on a unique feature of Australia’s history: convict
transportation from Britain. All the results are robust to instrumenting the overall sex
ratio by the sex ratio among convicts. Although convicts had no choice on where to
locate, they were not confined to prisons. They either worked under the government’s
supervision or were assigned to employers, who were either free settlers or former
convicts. As before, we remove the potential endogeneity that arises from this process of
convict assignment by controlling for the abovementioned geographic and historical
characteristics. Since a legacy of a convict past independent of gender imbalance would
violate the exclusion restriction, we also control for the number of convicts.
We undertake a number of additional robustness tests. In particular, we check that the
results do not rely on a specific measure of the sex ratio at one point in time. The results
are also robust to propensity score matching on the basis of geographic and historical
characteristics. Placebo specifications, in which sex ratios are randomly allocated across
counties while keeping the overall imbalance unchanged, give no significant results.
Moreover, we find no significant effects of the historic sex ratio on male work outcomes.
The results are specific to views about women working; historical gender imbalance does
not explain sexism in general, proxied by other survey questions on attitudes towards
women.
Our historical results, contemporaneous to gender imbalance, are in line with previous
literature. Economic and evolutionary biology models predict more conservative gender
roles as a result of male-biased sex ratios (Grossbard-Schechtman 1984, Chiappori 1988,
1992, Chiappori et al. 2002, Kokko and Jennions 2008). This is particularly relevant
when job opportunities for women are few or unattractive, as it was the case in 19th
century Australia, which was heavily specialized in the production of primary
commodities.
5
More surprising is that this effect has persisted to this day, after sex ratios have reverted
back to parity. To explain persistence, we argue that past gender imbalance has shaped
cultural beliefs about gender roles. We first rule out that other mechanisms explain
persistence. Our results rely on within-country and even within-state variation, where
formal legislation is identical. This rules out formal institutions as a persistence
mechanism.2 Another possibility is that past circumstances in the marriage market
influenced respective incentives of men and women to invest in education (Chiappori, et
al. 2009). We find no evidence for this mechanism. Initial gender imbalance could also
have distorted industrial specialization towards male-intensive economic activities. Since
we control for geographical endowments, such as land characteristics and mineral
discoveries, and for initial economic specialization, we view the remaining variation as
integrant to cultural persistence. We also analyze local historiographies and document
differences today between areas that had and still have a similar economic specialization
but different past sex ratios.
We next investigate the channels that underlie cultural persistence. Culture persists
because of the transmission of cultural traits within families or across unrelated
individuals (Cavalli-Sforza 1981, Bisin and Verdier 2001, Hauk and Saez-Marti 2002).
We find, consistent with vertical transmission, that historical gender imbalance is only
associated with conservative views about gender roles among people born of Australian
parents.
Our focus on the marriage market suggests an additional and novel – although not
necessarily the only other potential one – channel of cultural persistence. Assortative
mating in the marriage market makes gender norms strategic complements among
potential spouses. Strategic complementarity implies that norms become evolutionary
stable (Young 1998) so that even inferior conventions can persist over time, as shown
theoretically by Tabellini (2008) and Belloc and Bowles (2013). Accordingly, we find
2 The legal framework operating in Australia with respect to gender discrimination has been constant across all states since the Sex Discrimination Act 1984 (Cth), which operates at a federal level. This is a direct consequence of Australia’s Constitution, with any state law inconsistent with this act invalid to the extent of the inconsistency (Constitution s 109). The Family Law Act 1975 (Cth) unifies family law in Australia at this federal level.
6
that historical gender imbalance is only associated with conservative gender views in
areas where homogamy is high.
The main contributions of this paper are two-fold. The first is to shed light on the long-
term effects of gender imbalance. We show that the effects of gender imbalance on social
norms and on female occupations have persisted for more than a hundred years, even
though sex ratios have long reverted to normal. This has important implications for the
world today, where it has been estimated that a hundred million women are missing
(Hesketh and Xing 2006), namely in China, India, sub-Saharan Africa and the Caucasus.3
The study of the determinants of such gender imbalances has attracted a large literature.4
The study of its consequences is more limited because of evident reverse causality issues,
and we contribute to this by exploiting a unique natural experiment.
Our second contribution is to the literature on the influence of culture on economic
outcomes and, more precisely, on how culture emerges and persists. Until recently, the
rise in female labor force participation, the expansion of women’s economic and political
rights, as well as the reduction in fertility that has been observed in developed countries
were explained by technological change and the rise in returns to female labor.5 However,
several studies have also demonstrated how slow-changing cultural beliefs influence real
work choices, family formation and welfare.6 Regarding the origins of such beliefs,
Alesina et al. (2013) show that conservative gender norms stem from the introduction of
plough agriculture in pre-industrial societies. Our contribution is to illustrate a more rapid
cultural change, which took place within a homogenous population, and to document how
a large shock in the marriage market can shape cultural beliefs and have effects that
persist well into the long run.
3 See Anderson and Ray (2010) for sub-Saharan Africa and Brainerd (2013) for the Caucasus. 4 See Rao (1993), Hesketh and Xing (2006), Chung and Das Gupta (2007), Edlund and Lee (2009). 5 See Goldin and Katz (2002), Greenwood et al. (2005), Doepke and Tertilt (2009), Doepke et al. (2012), Olivetti (2013). 6 Fortin (2005) shows how gender role attitudes influence labor market outcomes. Alesina et al. (2013) establish a relationship between beliefs and participation of women in the economy and in politics. Bertrand et al. (2013) find that households in which women earn more than men are less likely to form and, once formed, are more likely to lead to divorce. Fernández (2008, 2013) and Fernández and Fogli (2009) show that preferences for fertility and for female labor force participation change slowly
7
2. Conceptual Background
In this section, we explain how a large shock to the marriage market, with a large excess
of males, can move social norms towards more conservative views of gender roles. We
also discuss how society can remain stuck in this new equilibrium, despite subsequent,
smaller changes in gender ratios.
Work by economists, social psychologists, and evolutionary biologists predict that a
male-biased sex ratio will result in conservative gender roles. In economic marriage
models, the bargaining position of one gender is proportional to its scarcity (Becker 1973,
1974). Accordingly, Pollet and Nettle (2008) find that the importance of men’s wealth for
marriage in the US at the beginning of the 20th century is positively correlated with local
sex ratios. Addressing the possible endogeneity between local marriage conditions and
local sex ratios, Abramitzky et al. (2011) exploit variation in World War I related deaths
in France. They find that a shortage of men is associated with men marrying more and
marrying up. In the case of Taiwan after the influx of the Chinese Nationalist Army in
1949, Francis (2011) shows that, conversely, a shortage of women leads to women
marrying more. An improvement in women’s bargaining position, resulting from higher
sex ratios, is also predicted to reduce female labor force participation (Grossbard-
Schechtman 1984, Chiappori 1988, 1992, Chiappori et al. 2002). This is supported by
empirical evidence based on the influence of contemporaneous changes in the sex ratios
of recent migrants on the outcomes of second-generation migrants (Angrist 2002).
Guttentag and Secord (1983) relate sex ratios to the status and roles of women. When
they are in the minority, the ability of women to use their bargaining power to gain
freedom and independence is limited if political and economic power resides in the hands
of men. In particular, the prediction is that in high sex ratio societies, women’s extra-
familial roles will be limited, although women may be treated with more respect and
greatly valued in their roles as homemakers (Guttentag and Secord 1983, p. 20).
While all this work puts forward bargaining as the main mechanism through which the
sex ratio affects gender roles, evolutionary biologists discuss others, including signaling
(Kokko and Jennions 2008). Male-biased sex ratios may also have negative consequences
8
for females when males divert resources from the female and offspring towards
competing with other males, or when males engage in mate guarding and restrict female
freedom. Regardless, the prediction applied to humans is still that male-biased sex ratios
will lead to more conservative gender roles and females working less outside the home.
As argued by Alesina et al. (2013), conservative gender roles and low female labor force
participation can imprint onto cultural norms the appropriate role of women in society.
Gender roles of the past then persist in the long run because culture changes slowly.
Culture is defined, after Richerson and Boyd (2005), as “rules of thumb” that affect
behavior in complex and uncertain environments and that people acquire from other
people through “teaching, imitation and other forms of social transmission” (Richerson
and Boyd 2005, p. 5). Cultural traits that are successful, which in our context means
getting a wife, will spread.
The economic literature discusses two main channels of cultural transmission: horizontal
and vertical (Cavalli-Sforza and Feldman 1981, Bisin and Verdier 2001). Culture spreads
horizontally across peers, mainly through imitation. Culture spreads vertically from
parents to children, through imitation and active parental socialization (Bisin and Verdier
2001, Doepke and Zilibotti 2008). Vertical transmission is inherently sticky, which
explains why historical sex ratios may have persistent effects, even long after sex ratios
have reverted back to normal.
Our focus on the marriage market suggests an additional and novel persistence
mechanism. Assortative mating in the marriage market implies that views about gender
roles are strategic complements among potential spouses. Individuals with similar
backgrounds and similar views are more likely to marry one another and more likely to
stay married (Becker et al. 1977, Lehrer-Chiswick 1993). In the marriage market, if
certain norms make matching more likely, and a match more successful, these norms will
prevail and persist over time. Conservative gender roles may thus persist in the long run
solely because they are mutual best responses in the marriage market. Positive feedbacks
of this kind are at the core of the persistence of cultural conventions in Belloc and Bowles
(2013), even when such conventions are Pareto-dominated. Young (1998) shows
theoretically that norms that are mutual best responses are evolutionary stable. Thus,
9
conservative cultural traits may persist even when they are no longer adaptive and even
long after sex ratios have reverted back to normal.
Belloc and Bowles (2013) discuss conditions that make the transition from one
convention to another more likely. Cultural change has the characteristics of a collective
action problem. The greater the cost of deviating from a given set of cultural traits and
the bigger the population size, the less likely it is that any cultural change will occur.
Deviation and experimentation may be particularly costly in the marriage market, where
time is of the essence, uncertainty substantial, and search costs relatively high. If holding
modern views leads to long delays in finding a spouse, people will conserve traditional
views. However, immigration should make experimentation easier and may accelerate
transition towards modern gender views. Conversely, homogamy in the marriage market,
a proxy for the strength of strategic complementarity of gender views, should be
associated with stronger persistence of norms.7 We test these predictions in the empirical
analysis that follows.
3. Historical Background, Data, and Results: Gender Imbalance, Female Work and
Marriage in 19th Century Australia
3.1. Historical Background
European settlement in Australia commenced after independence of the United States,
when it became the new destination of choice for the United Kingdom’s overflowing jail
population. Between 1787 and 1868, 132,308 and 24,960 convict men and women were
transported to Australia, mostly to Tasmania and New South Wales (hereafter, NSW),
which initially also included Queensland, the Australian Capital Territory, and Victoria
(Oxley 1996, p. 3). These convict men and women were not “hardened and professional
criminals” (Nicholas 1988, p. 3) but “ordinary working class men and women” (Nicholas
7 Homogamy interacts with vertical transmission in a very natural way, as discussed in Bisin et al. (2004). Parents want to instill in their children norms that will make them attractive in the marriage market. If they anticipate that the prevalence of conservative gender views is high among potential spouses, parents will try harder to transmit such views to their children. Data limitations prevent us from fully investigating this mechanism.
10
1988, p. 7). The majority of convicts were transported for property offences, such as petty
theft (Oxley 1996).
The extent of free migration to Australia was rather limited until the 1830s and the
gender imbalance was sustained by ongoing convict transportation for nearly a century.
Male convicts made up more than 80% of the adult male population of NSW in 1833.
Even among free migrants, men outnumbered women. It was mainly men who were
attracted to the economic opportunities offered in Australia, which consisted mainly of
pastoralism and mining, especially after the discovery of gold in the beginning of the
1850s. As can be seen in Figure 1, a male-biased sex ratio endured in Australia for more
than a century.
The settler population of Australia was ethnically homogenous. The vast majority of
convicts and free migrants came from England and Ireland. In the 1846 NSW Census,
90% of people born outside the colony came either from England or Ireland, with very
little heterogeneity across different localities within Australia.8
Essential to our identification strategy is to understand what determined the variation in
population and sex ratios across space. Upon arrival, convicts were not confined to prison
cells. Initially, they were assigned to work under government supervision. Later, as the
cost of caring for large numbers of convicts became too high, convicts were assigned for
private employment. Employers were government officials, free settlers, or ex-convicts,
since convicts were freed after the term of their sentence, generally 7 years. The
placement of convicts was dictated by labor requirements and decided in a highly
centralized way, as described by Governor Bligh of NSW in 1812:
“They (the convicts) were arranged in our book (…) in order to enable me to
distribute them according” (cited in Nicholas 1988, p. 15, emphasis added).
As for free settlers, their spatial distribution was determined also by economic
opportunities, namely in agriculture and mining, the two sectors in which Australia
specialized in the 19th century (McLean 2012).
8 50% came from England and 40% from Ireland. The standard deviation of the two distributions is only 0.05.
11
Male labor was at a high premium in the colonial economy. In 1816, Governor
Macquarie of NSW announced that male and female convicts must be paid £10 and £7
per annum, respectively (Nicholas 1988, p. 131). Meredith and Oxley (2005, p. 56)
document an even larger, 46%, gender pay gap in the non-convict population. One
explanation is that only men had the physical strength required for agricultural work and
building the country (Nicholas 1988). Alford (1984, p. 243) also suggests that the notion
that remaining within the home was a “woman’s proper place” already played a large role
in explaining why women were “undervalued and underemployed” in the labor market
(Nicholas 1988, p. 15).
Some convict women were confined to female factories, which were “a combination of
textile factory and female prison” (Salt 1984, p. 142) for women who had borne a child
out of wedlock, displeased their assigned master, or committed a crime. 9 Women worked
in female factories for a very low or no wage.10 Overall, Governor Macquarie of NSW
put it best when he stated that convict women had 3 choices: become a domestic servant,
live in a female factory, or marry (Alford 1984, p. 29). In the circumstances described
above, marriage seemed like the most attractive option. And the demand for wives was
high.
The authorities’ concern that “the disproportion of the sexes” would have “evil effects”
as men experienced “difficulty … in getting wives” (Select Committee on Transportation
1837-1838, p. xxvii) was well founded. Men were more than half as likely to be married
than women (see Table 1), who were under great pressure to be married. According to the
historical Census, more than 70% of women in Australia were married in the 19th century,
a much higher rate than in Britain at the same time period (60%) (Alford 1984, p. 26).
The legal ability to divorce came rather late, particularly in NSW (1873). By the end of
the 19th century (1892), there were a cumulated total of only 799 divorce petitions in
NSW (Golder 1985). Moreover, marriage was enforced by strict laws, as well as by
9 No analogous male factory existed. NSW had 3 female factories in the counties of Cumberland, Northumberland and Macquarie. Queensland’s county of Brisbane had 1 and Tasmania 5 (2 located in Hobart and the rest in Launceston, George Town and Campbell Town). 10 Third class women, those who committed a crime in the colony or misbehaved in the factory, received no wage (Salt 1984, pp. 86, 105).
12
Victorian morality. For example, bearing a child out of wedlock was considered a crime
for which women were sent to prison.
In sum, 19th century Australia was a context in which women’s economic opportunities
outside marriage were limited and unattractive. As a consequence, it is expected that
women would have been attracted to men who could fulfill the role of economic provider.
The high bargaining power of women, due to their scarcity, would further reduce their
incentives to work.
3.2. Historical Data
We collect data on the historical gender ratio and on the structure of the colonial
economy from the Colonial Censuses taken in the 19th century in each of the six
Australian states.11 Other data sources, such as colonial musters that counted transported
people, have high reporting error and are not representative of the entire population since
participation was not compulsory (Camm 1978, p. 112).
Our main measure of the historical sex ratio in regressions with present day outcomes is
taken from the first Census in each state. This is because we want to rely on the earliest
possible measure of the gender imbalance and of its exogenous component, which came
from convict transportation. We therefore rely mostly on the 1836 NSW Census (which
also included the Australian Capital Territory at the time), the 1842 Tasmanian Census,
the 1844 South Australian Census, the 1848 Western Australian Census, the 1854
Victorian Census, and the 1861 Queensland Census. These dates vary because states were
independent colonies until 1901.
Descriptive statistics for this historical cross-section are displayed in Panel A of Table 1.
Although the total population of Australia at the time was only about 255,000 people,12
more than 60% of the current population of Australia now lives in the areas that were
11 The online data from the Historical Census and Colonial Data Archive was supplemented by the actual Census report due to errors in the 1881 Tasmanian Census. Only the Census reports are available consistently across the period, as some of the individual records were destroyed in a fire in 1882. 12 These numbers do not include Aboriginal or Torres Strait Islanders, who were not counted in the Census until the 1960s. Only very rough estimates are available for these populations.
13
covered by the data. The unit of observation in the Census is a county.13 The average
county had 4,764 individuals, and the majority of counties (about 85%) had between 300
and 10,000 people. On average, the sex ratio stood at over 3 men for every woman in the
population but it was much higher among convicts, at nearly 30 men for every woman.
As an extreme example, in the county of Bligh, NSW, the sex ratio was 11 in the whole
population14 and reached 219 among convicts. The historical Census also contains
information on the number of married males and females and on economic occupations
by gender. Unfortunately, available records do not provide any further break down of
occupation by age or marital status.
Table 1 compares how well covariates are balanced between counties with historical sex
ratios above or below the median (2.05). Agriculture was the largest employer in
Australia at the time, accounting for 23% of the employed labor force. Next were
domestic services with 14%, and manufacturing and mining with a combined total of
12%. The shares of people employed in these different activities do not differ
systematically across high and low sex ratio areas. Areas with high or low historical sex
ratios are also broadly similar in terms of land characteristics and mineral endowments.
Areas with high historical sex ratios are richer in major gold deposits, but poorer in major
coal deposits. These statistics are displayed in Panel C of Table 1.
Figure 2 maps the sex ratio in the whole population and in the subset of the convict
population in areas of Australia that were already settled at the time of the study. The
population included in the historical Census came to Australia by sea. Yet, by the time we
measure them, people of both sexes had made their way into the hinterland and along the
coasts. The concentration of sexes has no definite pattern: high and low sex ratios were
found in the hinterland as well as along the coast.
For the historical regressions that follow, we also consider the full panel of 19th century
Censuses, roughly from 1836 to 1881, as described in Table A1. The panel is unbalanced
across states because of their status as independent colonies until 1901 and for some of
13 “Counties” will be used here to refer to historical administrative divisions in the different colonies of Australia, which were variously called “counties”, “police districts”, “towns”, or “districts”. 14 Hereon, “whole population” includes convicts, free settlers and emancipists (i.e. ex-convicts). Hence, the historical sex ratio is inclusive of the convict sex ratio.
14
the years the maps of counties are not available for some of the colonies. The panel stops
in 1881 because substantive redistricting occurred after that date and maps are not
available. Descriptive statistics for this panel dataset are included in Panel B of Table 1.
Under the balancing influence of natural births,15 the sex ratio over this whole period is
lower than in the first Census, but still stands at 1.9 men for every woman. Female
marriage rates were high throughout the period, particularly so in more male-biased areas.
There, while they married more, women worked less. In high sex ratio areas, female labor
force participation and the proportion of employed women in high-ranking occupations,
which include clerical, legal, and medical professions and teaching, are statistically
significantly lower.
3.3. Historical Regression Results
Panel estimates of the historical relationship between gender imbalance and marriage
rates, female labor force participation and the quality of female occupations are displayed
in Table 2. All specifications control for county fixed effects to remove the influence of
time invariant county characteristics that could be associated with gender imbalance and
marriage or female work outcomes. For each dependent variable, specifications in the
first column only include county fixed effects, and we add time fixed effects in the
second column. As the panel is unbalanced, we have grouped years together and consider
half decades as time fixed effects. Results are unchanged when we model time linearly
from the 1836 start date.
Gender imbalance is associated with higher marriage rates for women and lower marriage
rates for men. The effects are significant at the 1% level. More male-biased sex ratios are
also associated with lower female labor force participation and with a lower proportion of
women employed in high-ranking occupations. These effects are statistically significant
at the 1% level. They are also robust to controlling for the county’s male labor force
participation or for males employed in similar occupations. The results are large in
magnitude with an increase of one standard deviation in the sex ratio associated with a
15 Demographic studies of Australia have found no evidence of distorted or abnormal historical sex ratios at birth in Australia (Opeskin and Kippen 2012).
15
reduction in female labor force participation and in the share of women in high-ranking
occupations by 0.27 and 0.43 standard deviations, respectively.
All these results carry through in the historical cross-section provided by the first Census
in each state (see Table A2 in Appendix), even when controlling for geographical
characteristics (latitude, longitude, presence of minerals and land type), initial economic
specialization and state fixed effects.
To sum up, panel and cross section estimates indicate that in areas with higher gender
imbalance, women married more, worked less and were less likely to work in high-
ranking occupations.
In the next sections, we show how 19th century economic and marriage conditions have
shaped cultural traits and how they still influence outcomes in the present day.
4. The Legacy of Gender Imbalance on Culture and on Women in the Workplace
In this section, we explore the long-term consequences of gender imbalance for female
labor force participation and occupational choices and how it has shaped the cultural
values of Australians. First, we discuss how we link historical gender imbalance to
present-day opinion surveys and Census data.
4.1. Data
Postal areas in modern-day datasets are not equivalent to historical counties. Prior to this
study, digitized shapefiles on Australian historical Census boundaries did not exist. We
collected and digitized hard copies of maps from the National Library of Australia and
from State Libraries in order to construct these boundaries and match historical counties
to present-day boundaries. 16 The Appendix lists the maps used.
We explore the legacy of male-biased sex ratios on female labor force participation and
occupational choices with data from the most recent Australian Census, taken in 2011.
The unit of observation is the postal area. There are 2,515 postal areas in total. We match
16 When a postal area was found in multiple counties, we assigned it to the county in which it was mostly located.
16
slightly less than 2,000 of these postal areas to historical counties, the remaining 500 or
so being areas that were not settled at the time of the historical Censuses. To capture the
legacy of gender imbalance on female work choices, we focus on women in high-ranking
occupations: women employed as professionals as a proportion of the employed female
labor force.
Data on cultural attitudes today are from the Household, Income and Labor Dynamics in
Australia Survey (HILDA), a nationally representative survey available since 2001. The
location of respondents is identified by the postal area. After matching to historical data,
we are left with a sample of between 40,000 and 50,000 individual observations,
depending on the questions used, in more than 1,500 postcodes. Due to restricted survey
coverage, the number of postcodes is lower than that in the Census.
Questions on attitudes towards gender roles were included in the 2001, 2005, 2008 and
2011 waves. The main question that captures views about gender roles asks to what
extent respondents agree that: “it is better for everyone involved if the man earns the
money and the woman takes care of the home and children.” Response categories range
from 1 (strongly disagree) to 7 (strongly agree). We recoded this so that a higher value
indicates stronger disagreement with this statement, which we interpret as more
progressive attitudes.
We retain several individual characteristics from HILDA and variables from the Census
as well as data on mineral and land type from Geoscience Australia as controls.
Descriptive statistics are provided in Panels C and D of Table 1. The balance of these
covariates across areas below or above the median historical sex ratio is also presented in
the last two columns of Table 1. We observe no statistically significant difference across
high and low historical sex ratio areas in terms of age, gender, ancestry composition,
income, education, or gender balance today. We have already discussed how areas with
high or low historic sex ratio have similar endowments and land types. Areas that were
more imbalanced historically tend to be less urbanized today and, probably as a
consequence, marginally still more male biased. We include urbanization and the sex
ratio today as controls.
17
4.2. Specifications and OLS Results
Having matched historical gender imbalance to postal areas, we are able to examine its
legacy on attitudes, female labor force participation and occupational choices. Figure A1
in the Appendix shows that progressive attitudes towards gender roles, female labor force
participation, and the proportion of women employed in high-ranking occupations are all
negatively correlated with the historical sex ratio. The unconditional relationships
between all three outcomes and the historic sex ratio are all significant at the 1% level
and robust to the removal of outliers, such as areas with a sex ratio above 10. The simple
correlation coefficients between these variables stand well above the 0.1 mark considered
in Chatelain and Ralf (2014) as critical in regards to the possibility of spurious
regression.17
We explore the legacy of historical gender imbalance on present-day individual attitudes
and on female work by estimating the following equations:
(1) 𝑦!"# = 𝛼! + 𝛽!𝑆𝑒𝑥𝑅𝑎𝑡𝑖𝑜! + 𝑋!"! Γ! + 𝑋!!Π! + 𝑇!"! Λ! + 𝑋!"#! Θ! + 𝛿! + 𝛿! + 𝜀!"#
(2) 𝑦!" = 𝛼! + 𝛽!𝑆𝑒𝑥𝑅𝑎𝑡𝑖𝑜! + 𝑋!"! Γ! + 𝑋!!Π! + 𝑇!"! Λ! + 𝛿! + 𝜀!"
where 𝑦!"# is the survey-based measure of attitudes of individual i in postal area p, part of
historical county c. 𝑦!" are the Census-based measures of female labor force
participation or occupations in postal area p, part of historical county c. 𝑆𝑒𝑥𝑅𝑎𝑡𝑖𝑜! is the
historical sex ratio: the number of males over the number of females in historical county
c. 𝑋!"! is a vector of time-invariant geographic characteristics at the postal area level and
𝑋!! is a vector of historical controls. 𝑇!"! and 𝑋!"#! are vectors of postal area-level and
individual-level contemporary controls, respectively. 𝛿! is a vector of state dummies. 𝛿!
is a vector of HILDA wave dummies, when applicable. Since historical data at the level
of our 91 historical counties is less granular than present-day data at the postal area or
individual level, all standard errors are clustered at the county level.
17 The correlation coefficients between, on the one hand, the historical sex ratio and, on the other hand, progressive attitudes and female labor force participation are -0.25. The correlation coefficient between the historical sex ratio and the proportion of female professionals is -0.15. A correlation coefficient below 0.1 is considered by Chatelain and Ralf (2014) as too weak and possibly leading to spurious regressions. To avoid this issue, we also present regression results without the full battery of controls.
18
𝑋!"! and 𝑋!! are intended to capture geographic and historic characteristics that may have
been correlated with the sex ratio in the past and may still influence present-day
outcomes. In particular, as discussed in the preceding Section, Australia in the 19th
century specialized in the production of primary commodities in agriculture and mining.
Such economic opportunities influenced where convicts were assigned and where free
settlers located. If economic specialization persists over time for reasons separate from
the cultural channel we are interested in, these initial conditions could also influence
present-day economic opportunities for women and ignoring them would bias our
estimates. In order to flexibly account for geographic differences across counties that
may be correlated with agricultural potential, we control for latitude and longitude in all
specifications. To control more precisely for mining and agricultural opportunities, we
control for 9 detailed categories of mineral deposits18 and for land characteristics.19 We
do not include elevation because it shows very little variation, with 95% of our
population being in a low-grade area. We also control directly for the county historical
economic specialization, by including in 𝑋!! the historical shares of the population
employed in the main categories of employment discussed in Section 3: agriculture,
domestic services, mining and manufacturing, as well as in employment categories that
could provide opportunities for women: government and learned professions (including
teaching). Total historical population in the county is also included in 𝑋!!.
In the models of individual attitudes, present day individual controls include gender,
marital status, age, income, education, and whether the respondent was born in Australia.
Postal area-level controls include the sex ratio today and urbanization, taken from the
Census.
In the models of female labor force participation and occupational choice, contemporary
controls include the sex ratio today, urbanization and average education. Controlling for
the proportion of married people or for the full range of industrial specialization is
problematic, as these are endogenous outcomes. However, to account for sectorial
18 Minor coal; minor other; major coal; major copper; major gold; major mineral sands; major oil and gas; major others. The excluded category is no deposits or traces only. Source: Geoscience Australia. 19 Plains, plateaus and sand plains; hills and ridges; low plateaus and low hills; mountains. Source: Geoscience Australia.
19
differences across counties that influence the share of women employed as professionals,
we control for the share of men employed in similar occupations. Considering that we are
keeping the formal legislation constant by exploiting within-country and even within-
state variation, controlling for the share of men employed in similar occupations should
leave us with the variation that is due to culture, as opposed to formal institutions,
technology, or employment opportunities.
The estimates displayed in Table 3 show that where the gender imbalance was most
severe in the early days of colonial settlement in Australia, people are less likely to hold
progressive views about gender roles, and women are less likely to participate in the
labor force. When women do work, they are less likely to occupy high-ranking
occupations.
The relationship between attitudes towards gender roles and historical gender imbalance
remains statistically significant at the 1% level even when controlling for the full set of
geographic, historic, and contemporary controls. In terms of magnitude, the effect of the
historical sex ratio on attitudes is comparable, at the mean, to the effect of urbanization.20
Moreover, the relationship between the historical sex ratio and attitudes today is specific
to views about women working. Gender imbalance does not explain sexist attitudes in
general. For example, Table A3 includes the estimation results of equation (1) in which
the dependent variable captures respondents’ views about the quality of female leaders.
There is no significant relationship between historical gender imbalance and such
attitudes.
The relationship between female labor force participation and the historical sex ratio
remains negative but is no longer significant when we include all contemporary controls.
Female labor force participation may be too gross a measure as it pools together female
executives and check out chicks. When we focus on the quality of female work instead,
the relationship with historical gender imbalance remains statistically significant when
controlling for the full set of controls, including the share of men employed in similar
occupations. Additional results in Columns 4 and 5 of Table A3 in the Appendix show
that women are substituting full-time for part-time employment in high historical sex 20 The coefficient on the urban dummy, not reported, is 0.11 (s.e.: 0.08). The value of the historical sex ratio for the HILDA sample is 2.04.
20
ratios. A one standard deviation increase in the historical sex ratio is associated with a
decrease in the share of women employed as professionals by 0.13 standard deviations. In
terms of the share of the variation explained, adding historical characteristics to the full
set of controls increases the R-squared by 2.5 percentage points. This is equivalent to
more than 3% of the remaining unexplained variation in the share of women employed as
professionals.21
All the results pertaining to attitudes and to the share of women in high ranking
occupations are robust to non linear effects of the historical sex ratio and to excluding
metropolitan areas, counties that had fewer than 300 people or more than 40,000 people,
or counties that had very few women historically (less than 100). They are also robust to
controlling for distance to major ports of entry and to the main metropolitan areas, to
controlling for population density today, for the shares of different religions in the
population, historically and today22, or controlling for the average income in the county
and its quadratic. Some of these additional robustness tests are presented in Table A4. We
also check that the results are robust to propensity score matching. To do so, we predict
the historical sex ratio as a flexible function of extended geographic characteristics
(latitude, longitude, presence of minerals, land type) and historical employment shares in
different sectors as well as all interactions between geographic and historical
characteristics and second order polynomials. We then condition on this predicted
propensity score in the main specification. All the results described so far carry through
(see Columns 6 and 13 of Table A4). Columns 7 and 14 of Table A4 display the results
of placebo specifications in which historical sex ratios are randomly re-allocated between
historical counties, while keeping the overall average share of men relative to women
constant. As expected, the results are not significant.
Placebo specifications in which male work outcomes, such as male labor force
participation, the share of men employed as professionals and the share of men working
21 (0.573-0.559) / (1-0.559). 0.559 is the R-squared of a regression with all but historical controls. 22 Information on ethnicity is very sparse in the historical Census. Religion seems to be an adequate proxy. For example, in the above mentioned 1846 NSW Census, 50% of the population was Church of England, and 31% Roman Catholics, which corresponds well to the respective 50% and 40% shares of the population of English or Irish ancestry.
21
full- or part-time time are regressed on the historical sex ratio bear no significant results.
The results are in Table A5.
We implement a recent statistical test developed in Oster (2013). Based on the
recommended assumption that the maximum R-squared is 1.3 times the R-squared
obtained with the full set of controls, the influence of unobservable variables would need
to be 14 times as large as the influence of all controls included in Column 3 of Table 4 to
explain away the influence of the historical sex ratio for progressive attitudes. The
influence of unobservable variables would need to be nearly twice as large as the
influence of all controls included in Column 11 to explain away the influence of the
historical sex ratio for the share of women in high-ranking occupations.
We also check that the results are not idiosyncratic to a specific measure of the historical
sex ratio at a particular point in time. In Table A6, we present similar specifications as in
Table 3 but we rely on a later measure of the sex ratio, in 1861, by which date the total
population of Australia had increased nearly 5 fold. The 1861 sex ratio stood at an
average of 1.71. The specifications, set of controls, and results in Table A6 are identical
to those in Table 3.
4.3. Instrumental Variable Results
Our results are robust to a battery of observable geographic, historical, and contemporary
controls. Yet, where men and women chose to locate historically may have been driven in
part by unobservable characteristics, for example, on the basis of female preference for
leisure or male taste for discrimination. To address this concern, we adopt an instrumental
variable approach. We instrument the overall sex ratio by the sex ratio among the convict
population. The rationale for this instrumentation strategy is that convicts were not free to
choose where to go. Yet, convict assignment was not purely random and was influenced
by economic opportunities. As before, we remove this potential endogeneity by
controlling for the full set of geographic and historical employment sectorial shares. A
legacy of convict past independent of gender imbalance due, for example, to convicts
holding different views about gender roles than the rest of the population, would violate
22
the exclusion restriction. We therefore control for the number of convicts together with
historical population size.23
Table 4 presents the results of these instrumental variable specifications. Here we
regressed outcomes on the sex ratio instrumented by the convict sex ratio. These
specifications are estimated on a reduced population because convicts were only present
in 31 of our 91 historical counties; only in the states of Tasmania and NSW. In all
specifications, we include the full set of geographic, historical and present-day controls,
which are identical to those in Table 3. The first stages, for which statistics are displayed
in the lower panel of Table 4, are very strong. All the results discussed in Section 4.2 are
robust to this instrumental variable strategy.24
Because convicts were not present everywhere, the results in Tables 3 and 4 are not
directly comparable. Where convicts were present, the OLS coefficient corresponding to
Column 1 of Table 4 is -0.07 (s.e.: 0.019) and that corresponding to Column 2 is -0.25
(s.e.: 0.21). IV coefficients are larger in magnitude for two possible reasons. The first is
that the IV measures the local average treatment effect on the convict subpopulation. The
sex ratio was much higher in this subpopulation and while convict women married free
men, the reverse was rarely true. Convict women were therefore in a particularly well-
suited situation to extract a high bargain. The second is that the OLS coefficient may
suffer from attenuation bias. Convicts were all of marriage age, being between 15 and 50
years of age when transported (Nicholas et al. 1988, p.14), whereas the rest of the
population in the Census includes younger and older people. The convict sex ratio
therefore measures more precisely the sex ratio that is relevant for marriage and the
mechanisms we describe.
23 The presence of female factories, which hosted some female convicts, may have influenced the convict sex ratio as well as attitudes towards these women, who were considered outcasts. Controlling for the location of these factories does not alter our results and the effect of female factories is never significant. The potential endogeneity bias would run against the direction of our main result: we would find more conservative attitudes where there were more women. 24 We omit the results pertaining to female labor force participation, as these were neither significant in the OLS specification nor in the IV specification.
23
4.4. 1933 Results
We have so far documented the short-run implications of a male-biased sex ratio, and its
implications in the long run, about 150 years later. It is important for the validity of our
analysis to document medium-term implications, especially before the onset of massive
migration to Australia. Australia experienced its first significant influx of free migrants
after the discovery of gold in NSW and Victoria in the 1850s. However, deteriorating
economic conditions in the late 19th century and the White Australia Policy in the early
20th century restricted migratory flows (McLean 2012). The second period of mass
immigration into Australia occurred after the Second World War and the relaxation of the
White Australia Policy in the 1970s. In order to capture outcomes before these changes,
we rely on data on female work and occupations in the 1933 Census. We match 552 local
government areas (the unit of observation in the 1933 Census) to our historical counties
from the first Censuses. The total population of Australia in 1933 was 4.5 million people.
In 1933, the sex ratio still stood well above parity, at 1.16 (see Figure 1).
Figure A2 in the Appendix shows that the bivariate relationships between female labor
force participation or the proportion of women employed in high-ranking occupations in
1933 and the historical sex ratio are negative, statistically significant, and robust to the
removal of outliers. To check whether these relationships are robust to multivariate
analysis, we estimate specification (2) with female labor force participation and the share
of women employed in high-ranking occupations in 1933 as the dependent variables.
There is no urban/rural indicator in 1933, so we control instead for the share of people
employed in agriculture, in addition to tertiary education and to the sex ratio in 1933. As
before, we also control for male labor force participation or for the share of men in
similar occupations when relevant.
Regression results with the full set of controls are reported in Table A7. Female labor
force participation and the share of women in high-ranking occupations are negatively
associated with the historical sex ratio. The relationship remains statistically significant at
the 5% level for the quality of female work with the full set of controls.
To sum up, we have documented a large and persistent imprint of the historical sex ratio
on the share of women employed as professionals in the short, medium, and long run.
24
These differences in women’s employment outcomes are associated with persistent
differences in attitudes towards women working. The next section is devoted to studying
the underlying mechanisms of persistence, before turning to welfare implications in the
last section.
5. Cultural Transmission: The Roles of Vertical Transmission and of Marriage
Homogamy
We have documented in this paper a relatively rapid adaptation of cultural norms as a
response to gender imbalance, even among a population that was ethnically and culturally
homogenous to start with. These cultural norms have persisted over time, even after sex
ratios have reverted back to parity. Our focus on the marriage market suggests specific
cultural persistence mechanisms, which rely on homogamy in marriage and on child
socialization within families, which we explore in detail in this section.
If gender norms are transmitted within families, and if Australia’s past shaped a specific
norm in the way we describe, people whose parents are born in Australia should be more
likely to display this norm. To test for such vertical transmission in more detail, we add
interaction terms between historical gender imbalance and a dummy indicating at least
one Australian parent (mother or father).25 The excluded category consists of persons
born to two non-Australian born parents.
Regression results are in Columns 1 and 2 of Table 5. The coefficient associated with the
historical sex ratio alone is no longer significant, suggesting that historical gender
imbalance has no influence on people who are not born of Australian parents. The main
effect of having an Australian parent is positive and significant, but its interaction with
the historical sex ratio is negative, statistically significant and large in magnitude: more
than twice as large as in the sample as a whole. In other words, Australian parents
transmit more progressive norms, but not where the gender imbalance was high.
Consistent with the theoretical prediction regarding the role of migration in Belloc and
Bowles (2013) discussed in Section 2, we find that migration has an attenuating influence
25 Splitting this further into Australian born mother and Australian born father is problematic due to multicollinearity.
25
on cultural persistence. Belloc and Bowles (2013) discuss how immigration should make
experimentation easier and may accelerate transition from one cultural convention to
another. Consistent with this prediction, we find that the historical sex ratio is associated
with more conservative gender norms in low migration areas, but not in high migration
areas (Columns 3 and 4). These results also reveal that people of different ancestry
systematically display different attitudes although they live in the same area. This implies
that the relationship between historical imbalance and present-day outcomes is unlikely
to be due to unobservable local characteristics or to self-selection of people to localities
on the basis of taste.
We have discussed in Section 2 other mechanisms that might contribute to persistence
and in particular the issue of coordination on the marriage market and homogamy. We
document the relationship between homogamy in the marriage market and the persistence
of conservative gender norms in Columns 5 and 6 of Table 5. We define homogamy as
the probability that one’s partner is born in Australia when one is born in Australia, 86%
on average. Table A8 in the Appendix shows that homogamy brings direct benefits:
people are happier in their relationship when married to someone ethnically similar.
As we are concerned that homogamy is an endogenous outcome to the degree of cultural
persistence, we rely on a measure of homogamy predicted by several characteristics of
the postal area: education, the degree of urbanization, ancestry composition, average
median income, the sex ratio today and employment shares in 18 different sectors. We
define high and low homogamy postal areas as above or below the median predicted
homogamy.
The results show that historical gender imbalance is associated with conservative views
only in areas where predicted homogamy is high. By contrast, no legacy is found in areas
with low homogamy. Such strategic complementarities in the marriage market are
compatible with the apparent paradox of rapid adaptation of cultural norms yet cultural
persistence. The situation we study is that of a drastic shock to the marriage market, one
able to lead to rapid adaptation towards social norms that guarantee success in wooing a
wife. Our interpretation of persistence is that these norms locked in, even after sex ratios
had reverted back to normal, because of strategic complementarity of gender views in the
marriage market, together with vertical transmission within families.
26
In Appendix Table A9, we explore and rule out another transmission mechanism that
could explain persistence. Past circumstances in the marriage market may have durably
influenced respective incentives of men and women to invest in education (Chiappori et
al. 2009). To test for this, we regress the share of men and women with a tertiary
education in 2011 and in 1933 on the historical sex ratio, controlling for historical and
contemporaneous employment in different sectors of the economy, the usual geographic
controls and the contemporaneous sex ratio. The relationships are not statistically
significant.
Another possible channel is that initial gender imbalance distorted industrial
specialization towards male-intensive economic activities and that economic
specialization persisted over time. Since we control for initial economic specialization
and for geographical endowments, such as land characteristics and mineral discoveries,
we view the remaining variation as integrant to cultural persistence. Moreover, high and
low sex ratios areas actually did not differ systematically from one another in terms of
initial economic specialization, land characteristics, or mineral discoveries. We also
carefully analyzed local historiographies in order to contrast the fates of areas that had a
similar economic specialization in the past and are highly comparable on most observable
dimensions, but that had very different historic sex ratios. For example, county Bligh,
which we have already mentioned, and county Dalhousie are two inland counties
bordered by the Goulburn River and roughly equidistant from the nearest port. Both have
major coal deposits and consist mostly of low hills terrain. Both are rural counties that
were, and are still, predominantly specialized in agriculture. However, the sex ratio was
much more male biased in Bligh, with nearly 11 men for every woman against slightly
over 2 in Dalhousie. Today, in Bligh, female labor force participation is 47%, with 17%
of women employed as professionals, against 54% and 21% respectively in Dalhousie.
Our progressive attitude variable takes an average value of 2.05 in Bligh, against 3.78 in
Dalhousie.
27
6. Welfare Implications
Our historical results indicate that women were less likely to work and more likely to
marry in more male-biased areas. We have discussed how this may have been a
beneficial outcome for women given the options available to them in 19th century
Australia. We have documented negative implications of the historical sex ratio on the
quality of female occupations today, and on progressive attitudes towards female work.
Since conditions in the labor market have improved greatly for women since the 19th
century, these present-day outcomes may seem paradoxical in regards to the expected
enhanced bargaining power of women due to their historical scarcity. However, cultural
persistence implies that norms that were adaptive at a certain period of time may endure
even when they are no longer adaptive. Also, these outcomes are not direct indicators of
welfare. Therefore, we turn to more direct indicators of welfare in this section by
presenting evidence on the legacy of the historical sex ratio on wages and individual
satisfaction. Overall, we find no evidence that the effect on women’s welfare is negative.
6.1. Income
We have found that the historical sex ratio left an imprint on the quality of female
occupations. Here, we want to study other labor market outcomes, notably the gender
wage gap. In order to do so, we rely on reported individual wage information from
HILDA since the Census only records total income.
We estimate an equation similar to (1) with the log of financial year gross wages and
salary of individual i of gender g in postal area p in historical county c as the dependent
variable:
(3) 𝑙𝑛 𝑤!"#$ = 𝛼! + 𝛽!!𝑆𝑒𝑥𝑅𝑎𝑡𝑖𝑜! + 𝑋!"! Γ! + 𝑋!!Π! + 𝑇!"! Λ! + 𝑋!"#$! Θ! + 𝛿! + 𝛿! + 𝜀!"#
The vectors of time-invariant geographic county characteristics, historical controls, postal
area-level controls, and individual-level contemporary controls 𝑋!"! , 𝑋!! , 𝑇!"! and 𝑋!"#!
include the full set of usual controls. In addition, we control for the individual’s number
of hours worked. Differences with estimation equation (1) are that we now allow the
28
coefficient associated with the historical sex ratio, 𝛽!!, to be gender-specific. Standard
errors are, as usual, clustered at the county level.
We restrict our attention to full-time employees, between the ages of 18 and 65, with
non-zero reported wages.
Regression results are displayed in Columns 1 to 6 of Table 6. Column 1 includes only
controls for latitude and longitude, Column 2 adds the usual geographic, historical, postal
area and individual level controls. The coefficient associated with the main effect of
historical sex ratio is only marginally significant when the full set of controls is included.
Both the coefficients associated with female and with the interaction term between
female and the historical sex ratio are negative and statistically significant. In other
words, there is a gender gap in wage incomes, which is larger in areas that were more
male-biased in the past. Based on estimates in Column 2, moving from a county with a
historical sex ratio at parity (1) to a county with a historical sex ratio of 3 (the average) is
associated with a 6% increase in the wage earnings differential between men and women.
Moving from the county with the minimum sex ratio (1.01) to the county with the
maximum sex ratio (18.83) is associated with a 70% increase in the earnings differential
between genders.26
However, the earnings differential between genders that is explained by the historical sex
ratio is entirely accounted for by the number of hours worked and the fact that women
work fewer hours. In Columns 3 and 4, when we control for the number of hours worked,
the interaction between gender and the historical sex ratio is no longer statistically
significant. The earning differential has disappeared. Additional regressions in Columns 7
to 9, in which we regress the number of hours worked on the same set of controls as the
wage regressions, confirm that women in high historical sex ratios areas work fewer
hours.
As argued by Beaudry et al. (2012), industrial and occupational composition plays a large
influence on the structure of wages. We have flagged before the concern that industrial
specialization may be endogenous to the historical sex ratio, and including them as
26 These figures are computed as: 𝑒 !!.!"∗! − 𝑒 !!.!"∗! /𝑒 !!.!"∗! and 𝑒 !!.!"∗!".!" −𝑒 !!.!"∗!.!" /𝑒 !!.!"∗!".!" .
29
controls may bias our estimates, which is why we chose to focus on the estimates of (3).
Nevertheless, our results are robust to the inclusion of gender-specific industry (2*19
industry categories) and occupation dummies (2*8 occupation categories), which enable
a comparison of outcomes within industry and within occupation, as in Bidner and Sand
(2012). The occupation-gender dummies are included to account for the heterogeneity in
the quality of occupations between genders, which we have already documented. The
industry-gender specific effects are included to account for the possible heterogeneity in
income that could arise if men and women systematically sort into different industries
and if some industries systematically pay higher wages.27
The results of these specifications are in Columns 5 and 6 of Table 6. The results are
unchanged, except that the coefficient on female is no longer statistically significant once
we fully account for differences in industry and occupations between genders.
In order to address concerns about incidental truncation in the wage regressions, we
present in Appendix Table A10 the results of a Heckman correction for sample selection.
The excluded instruments in the selection equation consist of marital status and dividend
income. The null hypothesis of no selection bias is rejected in two thirds of the cases, but
all the results discussed so far are unchanged and similar in magnitude.
We have just reported that women work fewer hours and as a result earn lower income in
areas that had a high sex ratio historically, but this still says very little about women’s
welfare. To directly document this, we turn to self-reported satisfaction.
6.2. Satisfaction
We analyze self-reported satisfaction in two dimensions: marital satisfaction and overall
satisfaction. Marital satisfaction is captured in HILDA by the following question: “How
satisfied are you with your relationship with your partner?”, with answers ranging from
0 (completely dissatisfied) to 10 (completely satisfied). The sample average is 8.30, with
27 We have checked that all the results obtained with HILDA attitudinal data and discussed in Sections 4 and 5 are robust to controlling for the full set of gender-occupation and gender-industry dummies. For example, the coefficient with the full set of usual controls and these dummies, which corresponds to the coefficient in Column 3 of Table 3 is -0.026, with a t-stat of 4.12.
30
women less satisfied in their relationship than men (difference of -0.24, t-stat of 10.58).
Overall satisfaction is captured by the question: “How satisfied are you with your life?”,
with the same range of answers. The sample average is 7.91, with women more satisfied
than men (difference of 0.07, t-stat of 5.00).
We analyze answers to satisfaction questions as the dependent variable, and we contrast
the results for men and women. Regression results are in Table 7 and are first presented
with the restricted and then the full set of usual geographic, historic and contemporary
controls. We also add a control for the presence of small or dependent children.
Both men and women in high historic sex ratio areas are happier today in their
relationship. However, there is no significant difference between men and women; both
are happier. Yet, the relationship falls short of significance when the full set of controls is
added. Although the interaction between female and the historical sex ratio is positive, it
is not statistically different from zero.
People in high historic sex ratio areas are more satisfied in general. In contrast with
marital satisfaction, answers to the general life satisfaction question reveal significant and
robust differences between men and women as a function of the historical sex ratio. Even
though women are in general more satisfied than men, the relationship is reversed in
areas that had more men historically. In those areas, men are more satisfied than women.
Nevertheless, the overall effect of the historical sex ratio on female life satisfaction
remains positive and significant.
7. Conclusion
This paper documents the implications of missing women in the short, medium, and long
run. In areas with higher gender imbalance, women historically married more, worked
less, and were less likely to occupy high-rank occupations. In these areas today, people
have more conservative attitudes towards women working, women are still less likely to
have high-ranking occupations and they earn lower wage income because they work less.
Although our results may be specific to a certain technological context -work
opportunities for women were very poor in 19th century Australia- and although the
average deviation from a balanced sex ratio that we study in this paper is larger than
31
deviations observed today in certain parts of the world, a noteworthy implication is that a
temporary imbalance in the sex ratio has consequences that endure way beyond the
imbalance itself.
Our results illustrate how cultural norms emerged as a response to a specific scarcity
situation: the lack of women. Cultural norms can emerge as an adaptive evolutionary
response to a large shock in the marriage market and persist in the long run, when they
are no longer necessarily adaptive, despite subsequent but smaller changes to the
conditions in the marriage market.
We find that the presence of strategic complementarities between cultural norms, which
we discuss here in the context of the marriage market, underlies the persistence of culture.
We believe that the presence of strategic complementarity between cultural norms solves
the apparent paradox of rapid adaptation of cultural norms yet cultural persistence over
long periods of time. One implication is that persistence will be stronger and last longer
for norms that exhibit stronger strategic complementarities and in situations, like the
marriage market, where experimentation is costly. A more detailed exploration of this
mechanism is left for future research.
REFERENCES Abramitzky, Ran, Adeline Delavande, and Luis Vasconcelos. 2011. “Marrying Up: The
Role of Sex Ratio in Assortative Matching,” American Economic Journal: Applied Economics, 3(3): 124-157.
Alesina, Alberto, Paola Giuliano, and Nathan Nunn, 2013. “On the Origins of Gender Roles: Women and the Plough,” The Quarterly Journal of Economics, 128(2): 469-530.
Alford, Katrina. 1984. Production or Reproduction? An Economic History of Women in Australia, 1788-1850, Oxford University Press, Melbourne.
Almond, Douglas, Lena Edlund and Kevin Milligan. 2013. “Son Preference and the Persistence of Culture: Evidence from South and East Asian Immigrants to Canada,” The Population and Development Review, 39(1): 75-95.
Anderson, Siwan and Debraj Ray. 2010. “Missing Women: Age and Disease,” Review of Economic Studies, 77(4): 1262-1300.
Angrist, Josh. 2002. “How Do Sex Ratios Affect Marriage and Labor Markets? Evidence from America's Second Generation,” The Quarterly Journal of Economics, 117 (3): 997-1038.
Beaudry, Paul, David A. Green and Benjamin Sand. 2012. “Does Industrial Composition Matter for Wages? A Test of Search and Bargaining Theory,” Econometrica, 80 (3): 1063-1104.
Becker, Gary S. 1973. “A Theory of Marriage: Part I,” Journal of Political Economy, 81(4): 813–46.
Becker, Gary S. 1974. “A Theory of Marriage: Part II,” Journal of Political Economy, 82(2): S11–26.
32
Becker, Gary S., Elisabeth M. Landes, and Robert T. Michael, 1977. “An Economic Analysis of Marital Instability,” Journal of Political Economy, 85(6), 1141-1187.
Belloc, Marianna, and Samuel Bowles. 2013. “The Persistence of Inferior Cultural-Institutional Conventions,” American Economic Review, 103(3): 93-98
Bertrand, Marianne, Jessica Pan, and Emir Kamenica. 2013. “Gender Identity and Relative Income within Households,” NBER Working Paper 19023.
Bidner, Chris and Ben Sand. 2012. “Industrial Composition, Bargaining, and the Gender Wage Gap: Evidence from U.S. Cities,” Mimeo, York University.
Bisin, Alberto and Thierry Verdier. 2001. “The Economics of Cultural Transmission and the Dynamics of Preferences,” Journal of Economic Theory, 97(2): 298-319.
Bisin, Alberto, Giorgio Topa and Thierry Verdier. 2004. “Religious Intermarriage and Socialization in the United States,” Journal of Political Economy, 112(3): 615-664.
Brainerd, Elizabeth. 2013. “Missing Women in the Former Soviet Union? Son Preference and Children's Health in the Transition from Communism,” Working Paper, Brandeis University.
Camm, Jack. 1978, “Historical Geography and Past Populations: 19th Century New South Wales,” Australian Geographical Studies, 16(2): 111-125.
Carranza, Eliana. 2014. “Soil Endowments, Female Labor Force Participation, and the Demographic Deficit of Women in India.” American Economic Journal: Applied Economics, 6(4): 197-225.
Cavalli-Sforza, L. and Feldman, M. 1981. Cultural Transmission and Evolution: A Quantitative Approach. Princeton, NJ: Princeton University Press.
Chatelain, Jean-Bernard and Kirsten Ralf 2014. “Spurious Regressions and Near-Multicollinearity, with an Application to Aid, Policies and Growth,” Journal of Macroeconomics, 39(PA): 85-96.
Chiappori, Pierre-André. 1988. “Rational household labor supply,” Econometrica, 56(1): 63-90.
Chiappori, Pierre-André. 1992. “Collective Labor Supply and Welfare,” Journal of Political Economy, 100(3): 437-67.
Chiappori, Pierre-André, Bernard Fortin, and Guy Lacroix. 2002 “Marriage Market, Divorce Legislation, and Household Labor Supply,” Journal of Political Economy, 110 (1): 37-72.
Chiappori, Pierre-André, Murat Iyigun, and Yoram Weiss. 2009. “Investment in Schooling and the Marriage Market,” American Economic Review, 99(5): 1689-1713.
Chung, Woojin and Monica Das Gupta. 2007. “Why is Son Preference Declining in South Korea? The Role of Development and Public Policy, and the Implications for China and India,” Policy Research Working Paper 4373.
Commonwealth of Australia Constitution Act 1900 (Imp). Doepke, Matthias and Michele Tertilt. 2009. “Women’s liberation: What’s in it for men?,”
The Quarterly Journal of Economics, 124(4): 1541–91. Doepke, Matthias, Michele Tertilt, and Alessandra Voena. 2012. “The Economics and
Politics of Women’s Rights,” Annual Review of Economics, 4: 6.1-6.34. Doepke, Matthias and Fabrizio Zilibotti, 2008. “Occupational Choice and the Spirit of
Capitalism,” The Quarterly Journal of Economics, 123(2): 747-793. Edlund, Lena and Chulhee Lee. 2009. “Son Preference, Sex Selection and Economic
Development: Theory and Evidence from South Korea,” Columbia University Discussion Paper No. 0910-04.
Fernández, Raquel. 2008. “Culture and Economics,” in SN Durlauf & Blume LE (eds), The New Palgrave Dictionary of Economics, Palgrave Macmillan.
Fernández, Raquel. 2103. “Cultural Change as Learning: The Evolution of Female Labor Force Participation over a Century.” American Economic Review, 103(1): 472-500.
Fernández, Raquel and Alessandra Fogli. 2009. “Culture: An Empirical Investigation of Beliefs, Work and Fertility,” American Economic Journal: Macroeconomics, 1(1): 146-177.
Fortin, Nicole. 2005. “Gender Role Attitudes and the Labor-Market Outcomes of Women across OECD Countries,” Oxford Review of Economic Policy, 21(3): 416-438.
33
Francis, Andrew M. 2011. “Sex Ratios and the Red Dragon: Using the Chinese Communist Revolution to Explore the Effects of the Sex Ratio on Women and Children in Taiwan,” Journal of Population Economics, 24(3): 813-837.
Golder, Hilary. 1985. Divorce in 19th Century New South Wales. New South Wales University Press.
Goldin, Claudia, and Lawrence Katz. 2002. “The Power of the Pill: Oral Contraceptives and Women’s Career and Marriage Decisions,” Journal of Political Economy, 110(4): 730-770.
Greenwood, Jeremy, Ananth Seshadri, and Guillaume Vandenbroucke. 2005. “The Baby Boom and Baby Bust,” American Economic Review, 95(1): 183-207.
Grossbard-Shechtman, Amyra. 1984. “A Theory of Allocation of Time in Markets for Labour and Marriage,” The Economic Journal, 94: 863-882.
Guttentag, Marcia and Paul F. Secord. 1983. Too Many Women? The Sex Ratio Question. Beverly Hills, Calif: Sage.
Hauk, Esther and Maria Saez-Marti. 2002. “On the Cultural Transmission of Corruption,” Journal of Economic Theory, 107(2): 311-335.
Hesketh Therese and Zhu Wei Xing. 2006. “Abnormal Sex Ratios in Human Populations: Causes and Consequences,” Proceedings of the National Academy of Sciences of the U.S.A, 103 (36): 13271-13275.
Kokko, Hanna and Michael D. Jennions. 2008. “Parental Investment, Sexual Selection and Sex Ratios,” Journal of Evolutionary Biology, 21(4): 919-948.
Lehrer, Evelyn and Carmel Chiswick, 1993. “Religion as a Determinant of Marital Stability,” Demography, 30(3): 385-404.
McLean, Ian W. 2012. Why Australia Prospered: The Shifting Sources of Economic Growth. Princeton University Press.
Meredith, David and Deborah Oxley. 2005. “Contracting Convicts: The Convict Labour Market in Van Diemen’s Land 1840-1857,” Australian Economic History Review, 45(1): 45-72.
National Wage and Equal Pay Cases (1972) 147 CAR 172. Nicholas, S (eds) 1988, Convict Workers: Reinterpreting Australia’s Past, Cambridge
University Press, Australia. Olivetti, Claudia. 2013. “The Female Labor Force and Long-run Development: The
American Experience in Comparative Perspective,” NBER-Spencer Conference, Cambridge, Massachusetts.
Opeskin, Brian and Rebecca Kippen. 2012. “The Balance of the Sexes: The Feminisation of Australia’s Population, 1901-2008,” Population, Space, and Place, 18(5): 517-533.
Oster, Emily. 2013. “Unobservable Selection and Coefficient Stability: Theory and Evidence,” NBER Working Paper No. 19054.
Oxley, Deborah. 1996. Convict Maids: The Forced Migration of Women to Australia,. Cambridge University Press.
Pollet, Thomas V. and Daniel Nettle. 2008. “Driving a Hard Bargain: Sex Ratio and Male Marriage Success in a Historical U.S. Population,” Biology Letters 4:31-3
Rao, Vijayendra. 1993. “The Rising Price of Husbands: A Hedonic Analysis of Dowry Increases in Rural India,” Journal of Political Economy, 101(4): 666–677.
Richerson, Peter J. and Robert Boyd. 2005. Not By Genes Alone: How Culture Transformed Human Evolution. The University of Chicago Press.
Salt, Anette. 1984. These Outcast Women: The Parramatta female factory 1821-1848, Hale and Iremonger Pty Ltd, Sydney.
Select Committee on Transportation 1837–1838, Report, Minutes of Evidence, Appendix, Index, House of Commons Parliamentary Papers.
Qian, Nancy. 2008. “Missing Women and the Price of Tea in China: The Effect of Sex-Specific Earnings on Sex Imbalance,” The Quarterly Journal of Economics, 123 (3): 1251-1285.
Young, Peyton. 1998. Individual Strategies and Social Structure: An Evolutionary Theory of Institutions. Princeton: Princeton University Press.
34
TABLES Table 1: Summary statistics and balance of covariates
Variables Obs. Mean s.d. Min Max
Diff. above
vs. below
median historic sex ratio
(i)
t-stat of diff. (ii)
Panel A: First historical cross section for use in present-day regressions Historic sex ratio 93 3.10 2.95 1.01 18.83 3.11 6.04*** Convict gender ratio 36 28.86 42.04 1.27 219.00 32.9 3.55*** Historical Population (in thousands) 93 4493 11876 36 101080 -4,019 -1.63 Sectors – % of pop. employed in:
Agriculture, pastoral, horticulture or winegrowing 90 0.23 0.12 0.02 0.75 0.036 1.38
Domestic and personal service 90 0.14 0.14 0.03 0.81 -0.0094 -0.32 Manufacturing or Mining 90 0.12 0.20 0.00 1.23 -0.042 -0.95 Government and learned professions 90 0.02 0.02 0.00 0.13 -0.0042 -0.97
Panel B: Historical panel data (1836 - 1881) Historic sex ratio 446 1.88 1.57 0.93 18.83 1.23 8.63*** Prop. female married 406 72.37 26.60 6.26 284.36 12.3 6.05*** Prop. male married 406 39.53 12.93 5.34 138.58 -13.6 -8.75*** Female labour force particip. (% married women)
205 43.93 28.03 0 100.00 -20.6 -4.13***
Prop. women in high-ranking occupations 247 24.45 23.17 0 77.97 -10.3 -3.67*** Panel C: HILDA matched with the historical censuses and controls from 2011 Census and Geoscience Australia
Progressive Attitude Gender Roles 42,918 4.47 1.98 1 7 -0.37 -2.07** Log wages 24,263 10.52 0.92 0 13.93 0.00 0.07 Satisfied with partner 31,283 8.30 2.03 0 10 0.25 2.07** Overall life satisfaction 48,971 7.91 1.51 0 10 0.14 1.50 Individual Controls: Married or de facto 48,991 0.62 0.49 0 1 -0.02 -0.50 Age 49,019 43.69 18.42 14 101 1.33 0.94 Beyond year 12 education 48,989 0.34 0.47 0 1 -0.03 -0.55 Australia born 49,006 0.76 0.43 0 1 0.03 0.64 Male 49,019 0.47 0.50 0 1 0.02 0.64 Postal area controls: Urban 49,019 0.93 0.25 0 1 -0.20 -2.05** Contemporary gender ratio 49,017 0.97 0.09 0.64 13.53 0.04 2.10** Plains and plateaus 49,017 0.48 0.50 0 1 -0.13 -1.20 Hills and ridges 49,017 0.03 0.16 0 1 -0.11 -2.02** Low plateaus and low hills 49,017 0.06 0.23 0 1 0.07 1.82* Minor coal 49,017 0.04 0.20 0 1 -0.01 -0.87 Minor other 49,017 0.01 0.04 0 1 0 0.90 Major coal 49,017 0.27 0.44 0 1 -0.21 -2.86*** Major copper 49,017 0.01 0.05 0 1 0.05 1.68* Major gold 49,017 0.32 0.47 0 1 0.21 2.04** Major mineral sands 49,017 0.06 0.24 0 1 0.06 1.08 Major other 49,017 0.00 0.01 0 1 0.00 0.97
Panel D: 2011 Census matched to the historical censuses Female labor force participation 1900 56.00 9.47 0 100 -2.04 -2.27** Women in high-rank occupations 1890 21.31 8.48 0 46.50 -3.05 -3.29*** Additional Postal area controls: Prop. with professional college education 1895 0.21 0.05 0 1 -0.004 -0.87
Notes: (i) and (ii): differences and t-stat are from regressions at the county level with state fixed effects and robust s.e. In Panels A and B, observations are historical counties. In Panel C, observations are individuals. In Panel D, observations are present-day postal areas. The excluded land category is “mountains”. The excluded mineral category is “no traces or deposits”.
35
Table 2: Historical panel results: gender imbalance, female work and marriage between 1836 and 1881
1 2 3 4 5 6 7 8 9 10
Married women
(% women) Married men (% men) Female Labor Force Participation Women in high-ranking occupations (% working women)
Sex ratio 2.396*** 2.481*** -2.135*** -1.515*** -21.136*** -12.961** -13.205** -19.663*** -13.832*** -14.100***
(0.714) (0.794) (0.495) (0.285) (6.482) (5.076) (5.579) (3.452) (3.551) (3.603) Male labour force participation 0.088***
(0.026) Male high-rank occupation 0.110
(0.106)
Observations 412 412 412 412 205 205 205 247 247 247 R-squared 0.026 0.234 0.055 0.120 0.077 0.335 0.377 0.101 0.617 0.619 Number of counties 94 94 94 94 70 70 70 77 77 77 County FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Year FE No Yes No Yes No Yes Yes No Yes Yes Notes: The table reports OLS estimates. The unit of observation is a historic county-year (see Table A1 for years in each state). ‘Sex ratio’ is the number of men to the number of women. ‘Female (respectively, Male) Labor Force Participation’ is the proportion of females (respectively, male) employed, as a proportion of married females (respectively, males). ‘Women (respectively, Men) in high-rank occupations’ is the proportion of women (respectively, men) employed in ‘commerce and finance’, as a percentage of the employed females (respectively, males). See Table 1 for summary statistics. All regressions are with a constant. Robust standard errors are reported in parentheses. ***, ** and * indicate statistical significance at the 1%, 5% and 10% level, respectively.
36
Table 3: Gender imbalance, gender role attitudes and female work today: OLS results 1 2 3 4 5 6 7 8 9 10 11 Progressive attitude Female labor force participation Women in high-rank occupations Historical sex ratio -0.044*** -0.048*** -0.036*** -0.502*** -0.406*** -0.349** -0.116 -1.125*** -0.704*** -0.662*** -0.445*** (0.012) (0.016) (0.009) (0.156) (0.139) (0.171) (0.075) (0.268) (0.219) (0.149) (0.135) Male labour force participation
0.788*** (0.040)
Male high-rank occupation 0.394*** (0.051) State fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes HILDA wave fixed effects Yes Yes Yes - - - - - - - - Geographic controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Minerals and land type No Yes Yes No Yes Yes Yes No Yes Yes Yes Historic controls No Yes Yes No Yes Yes Yes No Yes Yes Yes Individual controls No No Yes - - - - - - - - Contemporary poa controls
No No Yes No No Yes Yes No No Yes Yes
Observations 42,916 42,334 42,284 1,898 1,871 1,862 1,862 1,888 1,861 1,861 1,861 Number of counties 81 78 78 91 88 88 88 91 88 88 88 R-squared 0.017 0.020 0.168 0.033 0.075 0.185 0.731 0.130 0.192 0.277 0.573 Notes: Columns 1 to 3 present results when the dependent variable is an individual’s response to the statement “it is better for everyone involved if the man earns the money and the woman takes care of the home and children.” Response categories range from 1 (strongly disagree) to 7 (strongly agree) (mean: 4.47), which we recoded so that a higher value indicates more progressive attitudes. The unit of observation is an individual in a postal area (hereafter, POA) in a historic county. ‘Geographic controls’ are a POA’s centroid’s latitude and longitude. ‘Minerals and land type’ is the presence and type of mineral deposit (minor coal; minor other; major coal; major copper; major gold; major mineral sands; major others) and land formation (plains and plateaus; hills and ridges; low plateaus and low hills; mountains), which are provided by Geoscience Australia. ‘Historic controls’ are: the historical county population, as well as the proportion of residents working historically in agriculture, domestic service, manufacturing, mining, government services and learned professions. ‘Individual controls’ are: gender, relationship status (married or de facto), age, whether one was born in Australia and whether one has education beyond year 12. These are all derived from HILDA. ‘Contemporary poa controls’ in Columns 1 to 3 are the number of men to women in a POA, whether POA is urban. ‘Contemporary poa controls’ in Columns 4 to 10 are the same as those in Columns 1 to 3 with an additional control of the average vocational tertiary education of a POA. Columns 4 to 6 present results when the dependent variable is the 2011 FLFP as reported in the 2011 Census. ‘Male labor force participation’ refers to the 2011 MLFP as reported in the 2011 Census. Columns 7 to 10 present results when the dependent variable is the proportion of employed women employed as professionals from the 2011 Census. ‘Men in high-rank occupations’ refers to the proportion of employed men employed as managers or professionals. The unit of observation in Columns 4 to 10 is a POA matched to its historic county. Standard errors are reported in parentheses and have been corrected for heteroskedasticity and for clustering at the county level. Results are robust to using robust standard errors. ***, ** and * indicate statistical significance at the 1%, 5% and 10% level, respectively.
37
Table 4: Gender imbalance and attitudes and female work today: IV results 1 2 Progressive attitude Women in high-rank
occupations Panel A: Second Stage IV estimation results Historical sex ratio -0.120*** -0.572* (0.037) (0.297) Geographic controls Yes Yes Minerals and land type Yes Yes Historic controls Yes Yes Individual controls Yes - Contemporary poa controls Yes Yes Number of convicts Yes Yes Men in high-rank occupations - Yes Observations 13,834 485 Number of counties 28 31 R-squared 0.166 0.815 State FE Yes Yes HILDA wave FE Yes - Panel B: First stage estimation results Convict sex ratio 0.032*** 0.037*** (0.001) (0.004) Observations 16,082 485 Number of counties 28 31 F statistic 2033.9 110.15 Adjusted R-squared 0.861 0.854
Notes: See Table 3 for the list of controls. The table reports 2SLS estimates. All the controls are identical to those included in Table 3 and are identical in the first and second stages (Panel A and B). Standard errors are reported in parentheses and have been corrected for heteroskedasticity and for clustering at the county level.
38
Table 5: Persistence: vertical cultural transmission and homogamy 1 2 3 4 5 6 Progressive attitude
Historical sex ratio 0.024 0.014 0.053** 0.038* 0.069 0.049 (0.018) (0.015) (0.026) (0.021) (0.047) (0.047) Australian parent 0.304*** 0.302*** (0.061) (0.060) Australian parent * sex ratio -0.060*** -0.059*** (0.022) (0.021) Low migration 0.101* 0.102 (0.058) (0.062) Low migration * sex ratio -0.078*** -0.074*** (0.025) (0.019) High homogamy 0.087 0.061 (0.112) (0.127) High homogamy * sex ratio -0.088* -0.080* (0.048) (0.049) Geographic controls Yes Yes Yes Yes Yes Yes Minerals and land type No Yes No Yes No Yes Historic controls No Yes No Yes No Yes Individual controls Yes Yes Yes Yes Yes Yes Contemporary poa controls Yes Yes Yes Yes Yes Yes Observations 42,866 42,284 42,866 42,284 42,947 42,947 R-squared 0.167 0.169 0.166 0.168 0.167 0.168 State FE Yes Yes Yes Yes Yes Yes HILDA wave FE Yes Yes Yes Yes Yes Yes Notes: See Table 3 for the list of controls. ‘Low migration’ refers to POAs where the proportion of residents born in Australia is higher than the mean proportion of residents born in Australia as per the 2011 Census. ‘Australian parent’: dummy equal one if respondent has an Australian father or an Australian mother (mean: 0.68). ‘High migration’ refers to POAs where the proportion of residents born in Australia is lower than the median proportion of residents born in Australia as per the 2011 Census. Homogamy refers to the average proportion of people of Australian descent in the POA who married someone also of Australian descent. ‘Low Homogamy’ are POAs whose predicted level of homogamy lies below the median level of homogamy, which is 86%. ‘High Homogamy’ refers to POAs whose predicted level of homogamy lies above the median. Homogamy is predicted by the sex ratio today, the degree of urbanisation, income, education, shares of employment in 18 different industries and respondents’ parents’ countries of birth in the POA. Standard errors are reported in parentheses and have been corrected for heteroskedasticity and for clustering at the county level.
39
Table 6: Welfare implications: wage income and hours worked
1 2 3 4 5 6 7 8 9 Log of gross financial year wages and salaries Hours worked per week
Historical sex ratio -0.016** -0.012* -0.030*** -0.027*** 0.001 -0.012*** 0.603*** 0.611*** 0.397*** (0.008) (0.007) (0.007) (0.006) (0.005) (0.005) (0.152) (0.140) (0.085) Female -0.425*** -0.407*** -0.118*** -0.116*** -0.232 0.122 -9.746*** -9.619*** -17.129*** (0.043) (0.039) (0.027) (0.024) (0.146) (0.133) (0.693) (0.682) (1.341) Female * Historical sex ratio -0.028* -0.030** -0.003 -0.006 -0.024** -0.009 -0.718*** -0.724*** -0.396*** (0.016) (0.014) (0.010) (0.008) (0.010) (0.007) (0.218) (0.209) (0.148) Hours worked per week 0.034*** 0.033*** 0.029*** (0.001) (0.001) (0.001) Geographic controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Minerals and land type No Yes No Yes Yes Yes No Yes Yes Historic controls No Yes No Yes Yes Yes No Yes Yes Individual controls No Yes No Yes Yes Yes No Yes Yes Contemporary poa controls No Yes No Yes Yes Yes No Yes Yes Occupation-gender dummies No No No No Yes Yes No No Yes Industry-gender dummies No No No No Yes Yes No No Yes Observations 24,275 23,936 24,275 23,936 23,810 23,810 28,840 28,465 28,313 R-squared 0.114 0.184 0.356 0.404 0.311 0.465 0.150 0.161 0.255 State FE Yes Yes Yes Yes Yes Yes Yes Yes Yes HILDA wave FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Notes: See Table 3 for a list of controls. The level of observation is an individual. The dependent variable in Columns 1 to 6 is the log of individual yearly gross wages and salary. ‘Occupation dummies’: managers; professionals; technician and trade workers; community and personal service workers; clerical and administrative workers; sales workers; machinery operators and drivers; laborers. ‘Industry dummies’: agriculture, forestry, fishing; mining; manufacturing; electricity, gas, water, waste; construction; wholesale trade; retail trade; accommodation and food; transport, postal and warehousing; information, media and telecommunications; financial and insurance; real estate; professional, scientific and technical; public administration and safety; education and training; health care and social; arts and recreation; other services. Occupation- and Industry-gender dummies are the full list of, respectively, occupation and industry dummies interacted with female. Standard errors are reported in parentheses and have been corrected for heteroskedasticity and for clustering at the county level.
40
Table 7: Welfare implications: satisfaction 1 2 3 4 Satisfied with partner Overall life satisfaction Historical sex ratio 0.039*** 0.017 0.045*** 0.027*** (0.010) (0.011) (0.009) (0.006) Female -0.245*** -0.253*** 0.093*** 0.088*** (0.029) (0.033) (0.019) (0.018) Female * Historical sex ratio 0.001 0.004 -0.012* -0.011* (0.012) (0.011) (0.006) (0.006) Geographic controls Yes Yes Yes Yes Minerals and land type No Yes No Yes Historic controls No Yes No Yes Individual controls No Yes No Yes Contemporary poa controls No Yes No Yes Family structure controls No Yes No Yes Observations 31,282 30,512 48,968 48,297 R-squared 0.009 0.063 0.003 0.019 State FE Yes Yes Yes Yes HILDA wave FE Yes Yes Yes Yes Notes: The level of observation is an individual. ‘Geographic controls’, ‘Minerals and land type’, ‘Historic controls’, ‘Contemporary poa controls’ and ‘Individual controls’ are as in Table 3. The dependent variable in Columns 1 and 2 is individual responses to the question: “how satisfied are you with your relationship with your partner?”. The dependent variable in Columns 3 and 4 is individual responses to the question: “how satisfied are you with your life?” Response categories to satisfaction questions range from 0 (completely dissatisfied) to 10 (completely satisfied). ‘Family structure’ indicates whether the household includes young children (under the age of 15) or dependent children. Standard errors are reported in parentheses and have been corrected for heteroskedasticity and for clustering at the county level.
41
FIGURES
Figure 1: Sex ratio in Australia: number of men to every woman, 1830-2002
Source: Australian Bureau of Statistics
11.5
22.5
3
1820 1840 1860 1880 1900 1920 1940 1960 1980 2000year
42
Figure 2: Gender imbalance in mid 19th century Australia: in the whole population (Left Panel) and among the subset of convicts (Right Panel)
Notes: The maps only show the parts of Australia for which Census data is available for the period of study. Left panel: Australian Capital Territory, New South
Wales, Queensland, South Australia, Tasmania, Victoria, and Western Australia. Right panel: Australian Capital Territory, New South Wales, Tasmania. Source: Australian Historical Census